Escherichia coli is both an important pathogen and a primary model organism for microbial biology. Nevertheless, about half of its proteins have not been characterized at any significant level of detail. This project aims at a comprehensive description of protein-protein interactions in E. coli by using exhaustive yeast two-hybrid screening. It will thus add significantly to ongoing attempts to characterize purified protein complexes by mass spectrometry (MS) and other methods. By contrast to the MS studies it will include direct binary interactions. In addition, this project focuses on protein domains of unknown proteins and map their interactions in more detail using both yeast two-hybrid screens and peptide array mapping. Most importantly, such mapping data will be used to identify biologically relevant and essential interactions by mutating specifically epitopes in vivo that are mediating these interactions. An array of cells with such interaction epitope mutants can then be tested for growth defects or more specific phenotypes under various conditions, thus showing which interactions are essential in general or only under certain conditions. [unreadable] The research design will include cloning of all open reading frames (ORFs) into two-hybrid bait and prey vectors and testing all pairwise combinations in an automated, array-based two-hybrid system. In addition to testing all full-length proteins, little-studied protein domains and cytoplasmic loops of transmembrane proteins will be searched for interactions, too. Finally, protein-domain interactions identified in these screens will be analyzed by testing the domain for binding to synthetic peptides from the partner protein. This will allow us to map binding sites to the amino acid level and subsequently study their atomic details using existing crystal or NMR structures. The end product of this project is a comprehensive molecular description of all protein-protein interactions in E. coli. Such maps can be used to integrate other data about metabolic reactions, gene regulation, and signalling networks to achieve a system level understanding of a bacterial cell. Eventually this information will allow us predict the behavior of a cell by computer modelling. With such a level of understanding, we will be both able to target defined processes therapeutically by antibiotics as well as modify the metabolism of the cell in order to use bacteria as chemical factories. [unreadable] [unreadable] [unreadable]